Prueba de evaluaciones de impacto algorítmicas

Las respuestas y explicaciones aparecen a medida que realiza el cuestionario.

1. What is the main purpose of an algorithmic impact assessment?
   • To measure how fast a model runs in production
   • To identify who a system could affect, possible harms and safeguards before and during deployment
   • To certify that a model has no errors
   • To calculate the licensing cost of AI software
2. Which kinds of assessments inspired the AIA approach?
   • Environmental and privacy impact assessments
   • Credit ratings and stock valuations
   • Medical board exams
   • Software speed benchmarks
3. In Canada's AIA, what does the questionnaire score determine?
   • The salary of the project lead
   • Whether the system may use cloud computing
   • One of four impact levels that trigger specific requirements
   • The number of users allowed
4. Where are Canada's completed AIAs made available?
   • They are kept secret by law
   • Only in internal audit files
   • On the government's open data portal
   • Only to the vendor who built the system
5. Under the EU AI Act, which article sets out the fundamental rights impact assessment?
   • Artículo 5
   • Article 27
   • Artículo 35
   • Article 99
6. Which deployer would clearly be expected to carry out a FRIA under the AI Act?
   • A gamer using a chatbot for fun
   • A startup training a spam filter for its own inbox
   • A bank deploying a high-risk credit scoring system
   • A student using an AI writing assistant
7. How does the FRIA relate to a GDPR data protection impact assessment?
   • It replaces the DPIA entirely
   • It complements the DPIA rather than replacing it
   • It is identical to the DPIA
   • It only applies when no personal data is used
8. According to the guide, what is a common misconception about AIAs?
   • That they involve affected groups
   • That they cover data sources
   • That they consider harms before deployment
   • That they are a one-time compliance form